Laurent Shafey
Software Engineer at Google
Mountain View, California, United States
Summary
🤩
Rockstar🎓
Top SchoolLaurent Shafey is a seasoned software engineer with 11 years at Google based in Mountain View, bringing deep expertise at the intersection of backend systems and machine learning. He has made notable open-source contributions to TensorFlow's Lingvo project, improving beam search flexibility, attention layers, and initialization strategies—work that reflects both production-grade engineering and ML model internals. Trained in rigorous European programs culminating in a PhD, he blends academic depth with practical system design and optimization. Colleagues can expect a developer who navigates complex ML codebases comfortably and surfaces clean, reusable abstractions that simplify downstream work.
11 years of coding experience
Master of Science (MS), Master of Science (MS) at École Supérieure d'Électricité / SupelecTechnischen Universität DarmstadtDoctor of Philosophy (Ph.D.), Doctor of Philosophy (Ph.D.) at Ecole polytechnique fédérale de LausanneLycée Pasteur - Classes Préparatoires aux Grandes Ecoles
Github Skills (10)
beam-search10
machine-learning10
lstm10
tensorflow10
python10
nlp9
machine-translation8
speech-recognition7
speech-to-text7
speech-synthesis7
Programming languages (1)
Github contributions (5)
tensorflow/lingvo
Feb 2020 - Oct 2022
Lingvo
Role in this project:
Back-end Developer & ML Engineer
Back-end Developer & ML EngineerContributions:4 reviews, 82 commits, 5 PRs in 2 years 7 months
Contributions summary:Laurent made significant contributions to the `tensorflow/lingvo` repository, focusing on enhancements to the beam search functionality. Their work involved making the `beam_search_helper` more flexible by decoupling it from specific encoder output content. Furthermore, they defined and implemented internal helper methods for the LSTMCellSimple, streamlining the code. The user also addressed initialization methods, incorporated stateless initialization techniques for variables, and fixed issues in the MultiHeadedAttention layer, indicating their proficiency in optimizing and improving machine learning model components.
asrtranslationctcspeech-recognitiontensorflow
google/paxml
Jun 2022 - Jan 2023
Pax is a Jax-based machine learning framework for training large scale models. Pax allows for advanced and fully configurable experimentation and parallelization, and has demonstrated industry leading model flop utilization rates.
Contributions:96 commits, 3 PRs, 9 pushes in 7 months
c4gptjaxlarge-language-modelsllm
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.